Modeling and simulation to assess the use of change in tumor size as primary endpoint in Phase II studies in oncology
L. Claret (1), V. Andre (2), D. de Alwis (2), R. Bruno (1)
(1) Pharsight Corp., Mountain View, USA, (2) Eli Lilly and Co, Erl Wood, UK
Objectives: To develop a model for progression-free-survival (PFS) in 2nd line NSCLC and assess the potential gain in efficiency in using change in tumor size (CTS) at first assessment (6-8 weeks) as opposed to PFS as a primary endpoint in Phase II studies of new oncology treatments using a simulation approach.
Methods: A model for PFS as a function of CTS and other prognostic factors was developed using data from the docetaxel arm (ECOG 0, 1 patients, n=240) of a Phase III study of docetaxel vs. pemetrexed in 2nd line NSCLC patients [1]. CTS was highly predictive of PFS (p<0.0001). The model was assessed using a visual predictive check. A randomized Phase II study of a new investigational treatment vs. docetaxel was simulated under various scenarios for the efficacy of the investigational treatment: from 0 to 100% increase in PFS over docetaxel (3.29 months). The PFS model was used to assess the CTS required to achieve the desired efficacy goals in term of PFS. Multiple replicates of study design scenario were simulated and study performance (% successful trials) was assessed to compare design and endpoints. A Log rank test was used to compare PFS and a t-test was used on the log ratio of tumor size at 1st assessment to baseline size [2].
Results: The power of a 120-patient randomized Phase II (2:1 randomization) was 60% based on PFS (40% increase i.e. 2.1 months) and 100% based on CTS. In all simulated scenarios, CTS was always more efficient than PFS (greater power with less patients). This gain in efficiency can be explained by the fact that CTS is based on continuous variable actual treatment effect on tumor size whereas PFS focuses on the time to progression and time to death, which is an indirect measure of treatment effect on tumor size. The use of CTS as the primary endpoint would allow to use randomized trials with smaller sample size and/or to detect smaller differences in PFS (e.g. the power to show a 1.3 month increase in PFS would be 93% using CTS vs. 34.2% using PFS).
Conclusions: There is a pressing need to improve Phase II clinical trial design in oncology in the hope to decrease the high failure rate in Phase III. The use of CTS as the primary endpoint in randomized Phase II studies as recently proposed by Karrison et al. [3] coupled with simulation models offer a powerful alternative to more traditional endpoints. A disease-specific survival model [4] can also be used to make inference on expected survival of the investigational treatment and to support go-no go decisions and Phase III study design [5].
References:
[1] Hanna N., Shepherd F.A., Fossella F.V. et al. Randomized phase III trial of pemetrexed versus docetaxel in patients with non-small-cell lung cancer previously treated with chemotherapy. J Clin Oncol, 22, 1589-1597, 2004.
[2] Lavin P.T. An alternative model for the evaluation of antitumor activity. Cancer Clin Trial. 4, 451-457, 1981.
[3] Karrison T.G., Maitland M.L., Stadler W.M. and Ratain M.J. J Natl Cancer Inst, 99, 1455-1461, 2007.
[4] Wang Y. et al, FDA Clin. Pharmacol. FDA Meeting of the Advisory Committee meeting, Rockville, March 18, 2008. http://www.fda.gov/ohrms/dockets/ac/cder08.html#PharmScience.
[5] Bruno R. and Claret L. On the use and value of drug-independent survival models to support clinical drug development in oncology. FDA Meeting of the Clinical Pharmacology Advisory Committee, Rockville, March 18, 2008
